木质素

Catalysis Today 85 (2003) 113–124 Understanding silica-supported metal catalysts: Pd/silica as a case study B.K. Min, A.K. Santra1, D.W. Goodman∗ Department of Chemistry, Texas A&M University, P.O. Box 30012, College Station, TX 77842-3012, USA Received 4 April 2003; received in revised form 20 May 2003; accepted 30 May 2003 Abstract Supported metal catalysts, particularly noble metals supported on SiO2, have attracted considerable attention due to the importance of the silica–metal interface in heterogeneous catalysis and in electronic device fabrication. Several important issues, e.g., the stability of the metal–oxide interface at working temperatures and pressures, are not well-understood. In this review, the present status of our understanding of the metal–silica interface is reviewed. Recent results of model studies in our laboratories on Pd/SiO2/Mo(1 1 2) using LEED, AES and STM are reported. In this work, epitaxial, ultrathin, well-ordered SiO2 films were grown on a Mo(1 1 2) substrate to circumvent complications that frequently arise from the silica–silicon interface present in silica thin films grown on silicon. © 2003 Published by Elsevier B.V. Keywords: Metal clusters; Supported catalysts; Oxide; Metal–support interaction; SMSI; STM; Palladium; Silica; Films; Inter-diffusion; Silicide; Alloying and sintering 1. Introduction The study of metal particles on oxide supports is of importance in heterogeneous catalysis because the size and nature of the interaction of a metal particle with an oxide support are critical in determining cat- alytic activity and selectivity [1–3]. It is well-known that metals on reducible oxides such as TiO2 [4,5] exhibit a strong metal–support interaction (SMSI). On the other hand, irreducible oxides like SiO2 are assumed to be relatively inert. However, in certain cases, silica has been shown to exhibit a metal–support ∗ Corresponding author. Tel.: +1-979-845-6822; fax: +1-979-845-0214. E-mail address: goodman@mail.chem.tamu.edu (D.W. Goodman). 1 Present address: Halliburton Energy Services, Duncan, OK, USA. interaction following a high temperature treatment [6–8]. Oxidation and reduction at elevated temperatures are essential steps for the preparation of supported, high surface area catalysts; however, these treatments can cause morphological changes of the dispersed metal particles arising from sintering and/or metal– support interactions. Therefore, it is of considerable importance to investigate and define optimal condi- tions for catalyst preparation, pretreatment and acti- vation [9]. Depending on the particular metal–oxide system, various morphological changes resulting from a metal–support interaction have been reported, namely sintering [10–14], encapsulation [15,16], inter-diffusion [17–27], and alloy formation [28,29]. In particular, silicide formation from metals supported on silica has received considerable attention be- cause of the importance of the metal–silica interface 0920-5861/$ – see front matter © 2003 Published by Elsevier B.V. doi:10.1016/S0920-5861(03)00380-8 114 B.K. Min et al. / Catalysis Today 85 (2003) 113–124 to numerous technologies. For example, studies of metal–oxide–semiconductor (MOS) structures are directly related to many aspects of semiconductor technology including the design of MOS devices. In addition, metallization is important for creating con- tact layers and durable electrically conducting vias on insulating substrates for semiconductor devices [30,31]. Furthermore, silicide formation between met- als and SiO2 in a catalyst has been shown to alter catalytic activity and selectivity [32,33]. For instance, it has been shown that Pd-silicide, formed during the high temperature reduction of Pd/SiO2, dramat- ically increases selectivity for the isomerization of neopentane [33]. However, in spite of the numerous studies on met- als supported on SiO2 at elevated temperatures, there are still controversial and unresolved issues regard- ing the nature of the metal–support interaction in silica-supported catalysts, namely: • the nature of the metal–support interaction between metals and SiO2; • the morphological changes that occur during the high temperature reduction of metals supported on SiO2; • the role of oxygen vacancies in the inter-diffusion of metals into SiO2; • the extent to which silicides are formed by the direct interaction between metals and SiO2; • the role of the silicon substrate, frequently used to prepare SiO2 thin films, in metal silicide formation; • the composition, if formed, of metal silicides; • the mechanism of silicide formation between metals and SiO2. In the first section of this review, the effect of high temperature reduction of technical catalysts consisting of metals supported on high surface area SiO2 will be discussed. In the second section, re- lated results obtained for model catalyst systems will be described. Finally, recent studies from our laboratories for model SiO2-supported Pd catalysts, consisting of SiO2 thin films prepared on a Mo sub- strate, will be addressed and serve to illustrate how this particular model preparation circumvents the complications frequently encountered when using a silicon substrate to synthesize a thin film silica support. 2. Experimental Details of the ultrahigh vacuum chamber, equipped with scanning tunneling microscope (STM), X-ray photoelectron spectroscopy (XPS), low energy elec- tron diffraction (LEED), and Auger electron spec- troscopy (AES) with a base pressure of 5×10−10 mbar, have been published elsewhere [34]. Briefly, this ap- paratus is equipped with a double-pass cylindrical mirror analyzer, reverse view LEED optics, and a room temperature STM (Omicron). Typically, the STM images were acquired in the constant current mode with a ∼2 V tip bias and a tunneling current of ∼0.1 nA. Ultrahigh purity (99.999%) oxygen from MG industries was used. A Mo(1 1 2) crystal, ori- ented with an accuracy of <0.25◦ (from Matek), was cleaned by flashing to 2100 K until no evidence of carbon and oxygen was detectable by AES. A W–5% Re/W–26% Re thermocouple was used to calibrate an optical pyrometer (OMEGA OS3700) that was then employed to monitor the crystal temperature during the STM experiments. 3. Results and discussion 3.1. High surface area, silica-supported metal catalysts Reduction at elevated temperature is one of the im- portant steps in the synthesis of supported metal cata- lysts; however, heating a supported metal catalyst can cause morphological changes in the metal particles depending upon the particular metal–oxide system. Chang et al. [9] have shown that various metal–support interactions are operative for Pd catalysts on vari- ous supports, e.g. SiO2, Al2O3, and TiO2, using a combination of temperature-programmed reduction and adsorption methods. Hydrogen adsorption exper- iments by this group on a SiO2-supported Pd catalyst (1% Pd/SiO2) showed no significant SMSI interaction even after heating to temperatures as high as 873 K. These results differ markedly from the data obtained for Pd/TiO2 where a significant reduction of adsorbed hydrogen was attributed to a strong metal–support interaction. The SMSI was attributed to arise via the diffusion of partially reduced TiO2 onto the surface of the Pd clusters. Based on these studies, the authors B.K. Min et al. / Catalysis Today 85 (2003) 113–124 115 suggested that the tendency for SMSI to occur be- tween dispersed Pd and a support increases in the order: Pd/SiO2 < Pd/Al2O3 < Pd/TiO2. In addition, it was also suggested that Pd/SiO2 displayed the least tendency to sinter compared with other metal–oxide pairs under identical reduction conditions. Negligible sintering of Pd/SiO2 (0.5–25.0% Pd) was reported by Moss et al. [35], who measured the Pd sur- face area by chemisorption of CO and H2 following two reduction temperatures, 573 and 723 K. Electron micrographs of the Pd/SiO2 catalysts corresponding to each reduction temperatures confirmed that no change has occurred in the metal morphology or dispersion; however, the specific catalytic activity for benzene hy- drogenation dramatically decreased with an increase in the reduction temperature to 823 K. Three expla- nations for this were proposed: (i) a variation in the surface structure of the support, (ii) promoters or in- hibitors on the Pd or the support, and (iii) an inter- action between the Pd clusters and the support. The latter explanation was based on an X-ray diffraction investigation that indicated formation of a Pd–silicon intermetallic compound. Direct evidence of Pd-silicide formation due to high temperature reduction has been suggested by Shen et al. [32,33]. Their XRD data of a mechanical mix- ture of Pd powder and silica gel after high tempera- ture reduction (873 K) showed the presence of a Pd3Si species. These authors suggested that the formation of this Pd-silicide species is responsible for the dra- matic increase in the selectivity of this catalyst for the isomerization of neopentane. The silicide formation is believed to occur during high temperature reduction by diffusion of Pd atoms into the bulk via oxygen va- cancies in the SiO2. Pure Pd, however, was easily re- generated by the oxidation of the silicide at low tem- perature. A change in catalytic activity following high tem- perature reduction has also been observed for Ni/SiO2 and Pt/SiO2. Martin et al. [36–39], have shown that the catalytic activity of Ni/SiO2 (24% of Ni) for ethane hydrogenolysis and benzene hydrogenation decreases by an order of magnitude with an increase in the reduc- tion temperature from 920 to 1120 K. Based on hydro- gen chemisorption it was anticipated that reduction at high temperature would lead to sintering of the nickel particles, however, the effects were much larger than that expected from a simple increase in particle size. A similar but more significant effect of high temperature reduction was also observed for a Pt/SiO2 catalyst. 3.2. Model silica-supported metal catalysts 3.2.1. Sintering Sintering [40–43] can be understood using the Gibbs–Thompson relationship where larger parti- cles with lower chemical potential will grow at the expense of smaller particles with higher chemical potential, the driving force being the reduction of the total surface energy of the system. Sintering is certainly affected by the environment and can be ac- celerated by a reactant gas and by temperature. In addition, sintering is strongly dependent on the na- ture of the metal–support interaction, i.e., the relative strengths of the metal–metal versus the metal–support bond energies. Generally, it is believed that for met- als on an irreducible oxide support, the strength of the metal–metal bond is significantly larger than the metal–support bond, leading to a relatively weak metal–support interaction and facile thermal sintering. Pretorius et al. [10] have investigated the interac- tion between various metals including noble metals and SiO2 thin films at high temperature (1073 K) using Rutherford backscattering (RBS) and scanning electron microscopy (SEM). Their results show that metals do not react with a SiO2 substrate whereas Ti, Zr, Hf, V, and Nb do react to form metal silicides. SEM images also have confirmed that islands of met- als tend to coalesce with a high temperature anneal. For noble metals, heats of formation were calculated and found to be positive using the following reaction: M + SiO2 → MSix + MOy. It was also suggested that these heats of formation should correlate with the mean electronegativity of the metal, allowing the rel- ative reactivity of a metal with SiO2 to be predicted. With these considerations, metals with an electroneg- ativity of less than 1.5 on the Pauling scale should react with a SiO2 substrate; the values for noble metals fall within the range 1.67–1.80, and therefore these metals should be unreactive [10]. Chen and Schmidt [11] have investigated the effects of reactive gases and temperature on the sintering rates and morphology of Pt particles on amorphous SiO2. Using transmission electron microscopy the average crystallite size of the Pt particles increased dramati- cally when heated to 973 K. 116 B.K. Min et al. / Catalysis Today 85 (2003) 113–124 The structure and chemical properties of model silica-supported metal catalysts have been investigated in our laboratory using various surface science tech- niques [12–14,44–46]. SiO2 thin films were prepared by high temperature co-deposition of Si and oxygen onto a Mo(1 1 0) or Mo(1 0 0) substrate followed by an anneal, a procedure that yields an amorphous ox- ide film. Temperature-programmed desorption (TPD) experiments for the Cu/SiO2 system showed that the saturation coverage of CO decreases by a factor of 2 with an increase in the annealing temperature from 100 to 900 K. In addition, marked changes in infrared reflection adsorption spectra (IRAS) were also ob- served. Fig. 1 shows IRAS of CO on silica-supported Cu as a function of the pre-anneal temperature. On the as-deposited (100 K) Cu clusters, CO exhibits an IR absorption band centered at 2099 cm−1 with a small shoulder on the low-frequency side of the peak. Upon annealing to 300–500 K the 2099 cm−1 Fig. 1. IRAS of CO on silica-supported Cu as a function of the pre-annealing temperature. Cu was deposited at 100 K and annealed to the indicated temperatures, followed by CO adsorption to saturation at 90 K. Fig. 2. TPD from Cu/SiO2/Mo(1 1 0). The silica film is ∼10 nm thick and the Cu coverage is ∼8× 1014 atoms/cm2 (∼0.6 MLE). The silica film was annealed to 1500 K before Cu deposition. Cu was deposited at 90 K and the heating rate for TPD was 10 K/s. band shifts to 2097 cm−1, and a new band appears at 2070 cm−1. Further heating to 700–900 K results in a splitting of the 2097 cm−1 feature into two peaks at 2018 and 2094 cm−1. These CO absorption bands are attributed to several distinct atop CO adsorption sites implying CO adsorption onto various Cu crystalline facets. TPD for CO adsorption at 90 K on Cu/SiO2 annealed to 1200 K shows no evidence for CO ad- sorption even though Cu is still evident on the surface as evidenced by the Cu TPD of Fig. 2. The �-peak at 1000 K was assigned to metallic Cu; the �-peak (1300 K) was attributed to a Cu species strongly bonded to the SiO2 surface since its desorption tem- perature was 200 K higher than the bulk Cu sublima- tion temperature. In addition, no adsorption of CO on Cu/SiO2, pre-annealed to 1200 K, indicated that this high temperature Cu species is not metallic in nature. The surface structure of supported Pd clusters on similarly prepared silica films was also investigated with IRAS of adsorbed CO. As shown in Fig. 3, for the as-deposited Pd film, two broad bands were ob- served at 2110 and 1990 cm−1, corresponding to CO adsorption on atop and bridging sites, respectively. Annealing the films to >500 K leads to a new band at 1890 cm−1, that corresponds to CO adsorbed on threefold hollow sites [14]. In addition, the peak for the atop and bridging bands becomes much sharper for the annealed films. This narrowing of the peaks is attributed to the formation of stable, extended facets. Fig. 4 shows the oxygen TPD from O/Pd/SiO2. B.K. Min et al. / Catalysis Today 85 (2003) 113–124 117 Fig. 3. IRAS of CO on a model silica-supported Pd catalyst (θPd = 15 MLE) as a function of pre-annealing temperature. The spectra were collected at 100 K and in 10−6 Torr CO background. The surface was annealed to 100, 300, 500, 700 and 900 K, respectively. Oxygen desorption occurs primarily within the tem- perature range 700–900 K, similar to oxygen desorp- tion from Pd single crystals and foils. In addition, a small oxygen desorption peak is observed between 1200 and 1300 K, the temperature range within which Pd is known to sublime. This high temperature oxygen Fig. 4. TPD from O/Pd/SiO2/Mo(1 1 0). The Pd coverage is 6× 1015 atoms/cm2. Oxygen was adsorbed at 100 K to saturation. desorption feature is attributed to oxygen absorbed within the bulk of the Pd clusters. However, oxygen desorption via decomposition of the silica film at high temperature catalyzed by Pd is also a possibility; this point will be discussed later in the text. 3.2.2. Encapsulation One deactivation mechanism for supported metal catalysts is encapsulation [1,15] resulting in a reduc- tion in the active metal surface area. This decrease in the ratio of the metal surface area to the metal in- terface area occurs when hemispherical metal clusters on an oxide surface are partially encapsulated by the oxide support. Powell and Whittington [15] first pro- posed encapsulation to account for the change in mor- phology of silica-supported platinum model catalysts following a high temperature anneal. These authors, using SEM, showed that Pt particles became partially immersed in the SiO2 surface with concomitant for- mation of a SiO2 ridge around the base of the Pt par- ticles when annealed at 1200 and 1375 K. Recently, Van den Oetelaar et al. [16,28] have in- vestigated the metal–support interaction between Cu and SiO2 using LEIS, AFM, and RBS by preparing very thick SiO2 films (400–500 nm) on a silicon wafer. The LEIS measurements showed that the coverage of Cu gradually decreased following an anneal to 773 K and completely disappeared after an anneal to 923 K indicating that no bare Cu remained on the surface. The RBS data, however, were unchanged following an anneal to 923 K, consistent with encapsulation of the Cu clusters. 3.2.3. Inter-diffusion Numerous investigations have shown that metals can diffuse into a SiO2 support. Scott and Lau [17] have investigated the effects of interfacial SiO2 on the formation of a metal silicide with a silicon substrate using RBS and 16O(d,�)14N/18O(d,�)15N nuclear reactions. A SiO2 layer was synthesized on a silicon wafer by oxidation, then a noble metal, e.g. Pd, Pt, Ni, evaporated onto it. A feature corresponding to a metal silicide was observed at the interfacial region of SiO2 after an anneal to 673–1023 K. As a plausible explanation, the authors suggested the following. Ini- tially, the SiO2 is present as a layer between the metal and the silicon. Upon annealing at a sufficiently high temperature, the metal diffuses through the SiO2 layer 118 B.K. Min et al. / Catalysis Today 85 (2003) 113–124 to form a silicide. The formation of the silicide gives rise to lateral non-uniformities in the layer such that the barrier to silicon diffusion is significantly lowered, i.e., silicon diffusion, no longer blocked by the SiO2 layer, is facilitated. Schleich et al. [18] have also investigated Pd-silicide formation on oxidized Si(1 1 1) using XPS and HREELS. These authors showed that a loss feature related to a C–O vibration near 240 mV disappeared after annealing to >510 K implying the disappearance of Pd on SiO2. Furthermore, subsequent to a 820 K an- neal, the HREELS spectrum showed a feature similar to that acquired from a freshly prepared SiO2 surface. In addition, XPS spectra showed an additional feature after a 600 K anneal near 336.9 eV in addition to the feature related to Pd0. This additional feature was consistent with the formation of Pd-silicide; silicide formation was confirmed by XPS spectra following an anneal of Pd on a clean Si(1 1 1)-7 × 7 surface. Based on these results these authors concluded that annealing leads to diffusion of Pd to the Si/SiO2 in- terface, likely via oxygen vacancies. At the interface, elemental silicon is available for reaction with Pd to form Pd2Si. Anton et al. [19,20] have investigated the growth of Pd on thermally grown or native silica layers at high temperature using AES, reflection high energy elec- tron diffraction, SEM, and Auger sputter profiling. In this study, the decrease of the metal AES intensity following an anneal to 893 K was interpreted as due to diffusion of metal clusters into the oxide la